For dealing with the green two-echelon multi-period vehicle routing problem with flexible time windows (G2E-MPVRPFTW), this paper establishes a mathematical model with the objectives of minimizing the carbon emissions and maximizing the customer satisfaction, and proposes a hyper-heuristic ant colony optimization algorithm (HHACOA) which combines the K-means clustering with time windows (KCTW). Firstly, according to the complex characteristics of G2E-MPVRPFTW with the large scale, multi constraints, and strong coupling, KCTW is adopted to decompose the problem into multiple subproblems. Thereby, the complexity of solving the problem is reduced. Secondly, HHACOA is used to solve the decomposed subproblems, and the solution of the original problem G2E-MPVRPFTW can be obtained by merging the solutions of these subproblems. In the policy domain of the upper layer, HHACOA generates different permutations of 9 neighborhood operations, and uses ant colony optimization algorithm (ACOA) to learn high-quality permutation information. Based on the reconstructed transition probability matrix, new permutations are generated to effectively guide the search to reach areas where the high-quality solutions are concentrated. In the problem domain of the lower layer, HHACOA utilizes the heuristic rules and the random method to generate the initial population, and uses each permutation generated at the upper layer as an algorithm to act on each individual in the population, so as to search more different regions in the solution space.